[openstack-dev] Dependency version pinning [was Re: Pyparsing 2.0 breaking jenkins jobs]

Sean Dague sdague at linux.vnet.ibm.com
Fri Mar 1 15:07:06 UTC 2013


On 03/01/2013 05:09 AM, Mark McLoughlin wrote:
> On Fri, 2013-03-01 at 20:27 +1030, Christopher Yeoh wrote:
>> On Thu, 28 Feb 2013 13:16:23 +0000
>> Mark McLoughlin <markmc at redhat.com> wrote:
>>>> That way we find out about compatibility problems ASAP but we have a
>>>> controlled way of managing the changes necessary into projects and
>>>> not end up with the gate stalled for hours at random times.
>>>
>>> It's not just about the gate, it's also about the software we ship.
>>> The ideal is that all our dependencies listed in setup.py are >=
>>> because that allows our software to be installed in more environments.
>>>
>>
>> That's true, though there's no reason we can't in general ship with >=
>> dependencies, but have the gate test with specific versions required,
>> where that version is the latest we know that works (determined through
>> some automated testing to warn us of problems ahead).
>
> I'd like us to be testing with two dependency stacks - both the minimum
> required versions and the latest available versions - but that's a ways
> away yet.
>
> The instability in the gate is best fixed by only updating the pypi
> mirror when the update doesn't break the gate. If a pypi mirror update
> fails, that means something is broken upstream of us and we either need
> to get the upstream fixed or pin to a specific version.
>
> Advantage is we know when something has broken and we can investigate it
> without the "OMG, the gate is hosed!" pressure.

Today the mirror isn't strictly enforced. Is there someone working on 
changing that in the near term?

	-Sean

-- 
Sean Dague
IBM Linux Technology Center
email: sdague at linux.vnet.ibm.com
alt-email: sldague at us.ibm.com




More information about the OpenStack-dev mailing list